SpeaKoach

High-Stakes Setup. Our Fusion Architecture adapts to your reality. Upload your CV, define the company (Big C), and the role (Cleaner).
SpeaKoach The AI identifies significant safety gaps (PPE, spill kits) in Jakkarin's answer. Objective feedback you can trust

Inspiration

Our project was born from a common human anxiety: the sweaty palms and frozen feeling before any crucial communication, whether it's a presentation or a job interview. We realized that success and career trajectories often hinge on good speaking skills. The question that drove us was personal: "Could improving my presentation skills fundamentally change my life's path?"

We identified a major gap: ** Current practice tools are uni-dimensional** , focusing on either speech or body language, but never both simultaneously. Real-world communication demands the integration of both. Our goal became clear: to build an all-in-one, objective, 24/7 platform to help everyone master communication and finally beat that anxiety.

What it does

SpeaKoach is an AI Communication Coach that helps you improve speaking skills in real-world situations. Unlike other tools that only focus on either speech or body language, SpeaKoach provides a complete practice environment with feedback on three important communication areas:

Non-Verbal (Computer Vision): Monitors your eye contact, posture, and any distracting movements like fidgeting, all in real-time.
Contextual Coaching (LLM): Uses AI to create realistic role-play scenarios and gives personalized feedback on your overall performance.

Key Modes: Interview Mastery: Practice with AI interviewers that ask intelligent follow-up questions based on your previous answers.

Presentation Coach: Receive real-time feedback on your content organization and delivery.

How we built it

We powered SpeaKoach using the OpenAI API as the core intelligence engine for a highly personalized, real-time simulator.

Input-Driven Customization: Our program consumes the user's Resume/CV to generate dynamic, context-aware interview questions using the LLM.
Real-Time Interaction: We integrated Speech-to-Text (STT) for real-time voice capture. This allowed the AI to instantly analyze the content of the user's spoken response and drive a fluid, natural conversational flow.

Challenges we ran into

Our core challenge was achieving data pipeline integrity and low latency under the constraints of real-time multi-modal processing:

Audio Overwrite Issues: The system struggled with concurrent audio file processing. We frequently ran into bugs where a new audio response would attempt to overwrite a file that was still being transcribed. This required implementing precise logic for sequential and safe handling of every audio chunk.
Minimizing Latency : Optimizing the entire sequential pipeline (Recording $\rightarrow$ STT $\rightarrow$ LLM Analysis $\rightarrow$ AI Response) was critical to ensure the simulated interview felt immediate and smooth.

Accomplishments that we're proud of

1.Achieving True Multi-Modality: We successfully built a platform that fuses video and audio inputs simultaneously, providing a comprehensive understanding of a user’s communication style—a significant technological leap beyond conventional tools.

2.The Dynamic Interview Engine: Our AI interviewer moves past scripted Q&A. It uses the LLM to analyze the user's content and generate complex, relevant follow-up questions, making the practice incredibly realistic and challenging

3.Objective Gaze Tracking: We implemented robust, non-intrusive eye-contact tracking, providing the objective data needed to conquer a primary physical manifestation of communication anxiety.

What we learned

The core lesson was in accelerated, practical system design:

-We learned the critical importance of data pipeline integrity when dealing with real-time audio streams.

-We mastered rapid iteration to fix complex bugs (like the audio overwrite issue) under tight deadlines.

We understood that the true power of the LLM lies in contextual prompting —making the AI act like a human coach, not just a script generator

What's next for SpeaKoach

Our path forward is clear. First, Monetization: We launch a Freemium Pro Tier, targeting B2B sales to HR departments next. Second, Feature Expansion: We will integrate Emotional AI and Vocal Tone analysis for deeper feedback. And third, Market Growth: We focus on University Career Centers and ambitious professionals. SpeaKoach is ready to scale.